knockouts-redesign
About
This is the implementation of a data-driven approach for discovering improvement opportunities related to the execution of knock-out checks within business processes.
The tool comes with a browser-based UI (powered by Streamlit), but it can also be invoked via CLI or imported as a module in Python scripts.
Resources
- Download the full paper here
- Download the full thesis here
- Event logs used for the paper experiments
- Results of baseline experiments
Index
Set-up
Option A: Using a setup script
- Setup scripts are provided for Windows & Linux, which perform the steps described in the "Manual steps" section.
- Requirements:
- Having cloned the repository
- Python 3.9+ installed
-
Commands:
# Windows: setup.bat # Linux: bash setup.shOption B: Manual steps
Expand
- Clone the repo - [Create a virtual environment with venv](https://packaging.python.org/en/latest/guides/installing-using-pip-and-virtual-environments/#creating-a-virtual-environment) - In the directory that contains the newly created `venv` folder, run: ``` source venv/Scripts/activate pip install -r requirements.txt ``` - Depending on your platform, install graphviz (v 3.0.0+) separately, and make sure it's in your `PATH` (see also: [graphviz downloads page](https://graphviz.org/download/#windows)): ```bash # Windows: choco install graphviz # Linux: sudo apt install graphviz # Mac: brew install graphviz ```Usage
Create a config file following the schema defined in config_schema.json (the directory also contains examples). Then you can choose whether to:
-
Launch the tool as a web app (powered by Streamlit):
# Windows: start.bat # Linux: bash start.sh -
Launch from CLI:
# Windows: python main.py config/synthetic_example.json # Linux: python3 main.py config/synthetic_example.json -
Or programmatically:
from knockout_ios.pipeline_wrapper import Pipeline ko_redesign_adviser = Pipeline(config_file_name="synthetic_example.json").run_pipeline()In all these cases, if using the default settings, output will be written to
.csvandtxtfiles in thedata/outputsfolder.
Running tests
Shell scripts are provided, which run the tests in parallel (thanks to pytest-xdist).
# Windows:
test.bat
# Linux:
bash test.sh